2010 AEG Outstanding Student Professional Paper

Transcription

1 2010 AEG Outstanding Student Professional Paper KEVIN T. MININGER RJH Consultants Inc., 9800 Mt. Pyramid Court, Suite 330, Englewood, CO PAUL M. SANTI Department of Geological Engineering, Colorado School of Mines, 1516 Illinois Street, Golden, CO RICHARD D. SHORT Slope Reinforcement Technology, LLC, 4125 Blackhawk Plaza Circle, Danville, CA Kevin Mininger was born and raised in Sod, West Virginia. After high school, he attended Guilford College in Greensboro, North Carolina, where he received a Bachelor of Science joint degree in Geology and Environmental Studies. He then moved to Chile for 3 years, two of which were spent working as a geologist in a small underground copper mine. After returning to the United States, he moved to the Denver area to attend the Colorado School of Mines. While working toward his Masters degree, he worked with Dr. Paul Santi, studying the factors that affect the clogging of horizontal wick drains and the ways in which these factors influence the service life span of the drains. In December 2010, he graduated with a Master of Science degree in Geological Engineering. Since June of 2010, Kevin has been employed by RJH Consultants, Inc., a geotechnical and water resources engineering firm in Englewood, Colorado. His work at RJH Consultants has included site investigations, feasibility studies, and preliminary design of dams and pipelines in Colorado, Wyoming, and Utah. Life Span of Horizontal Wick Drains Used for Landslide Drainage Key Terms: Horizontal Drains, Wick Drains, Clogging Rate, Landslides, Stabilization ABSTRACT Horizontal drains have been used since the 1930s to increase the stability of slopes by lowering the groundwater table. Horizontal wick drains represent a new technique that has proven to be effective at stabilizing slopes of fine-grained soils while being less costly and time-consuming to install than traditional polyvinyl chloride (PVC) or steel pipe drains. Their adoption has been slow in part due to uncertainty about clogging and the life span before drain replacement is required. To address these concerns, the degree of clogging and the reduction in flow rate were measured for exhumed wick drains that had been in the ground between less than 1 year and 11 years. The flow rate through the drain samples was measured under constant head conditions, and clogging was measured visually and by increase in drain mass. The results of this study suggest that drain failure does not occur during the first 11 years of use and is not anticipated during continued use in the expected field conditions. A reduction in flow rate of 85 percent was the maximum measured for the drain samples. Despite this reduction in flow rate, horizontal wick drains are able to transport at least two times more water than the surrounding soil will introduce. Though horizontal wick drains are not expected to fail due to clogging, a chart is provided to select the proper drain spacing based on soil hydraulic conductivity and required drain length, in order to prevent the flow capacity of the drains from being exceeded. Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

2 Mininger, Santi, and Short INTRODUCTION AND HISTORY Horizontal drains have been used for decades to increase the stability of slopes by lowering the groundwater table. They allow an easy path for water to drain from the slope under the influence of gravity. The first horizontal drains were installed in 1939 by the California Division of Highways (Smith and Stafford, 1957). These drains were installed by drilling a hole, angled slightly upward, into the slope and then installing a perforated steel pipe. With time, the method was improved, allowing the pipe to be inserted inside the drill stem, which eliminated problems from borehole collapse (Royster, 1980). The drain pipe also changed with the introduction of polyvinyl chloride, or PVC. PVC is less vulnerable to corrosion than steel and allows for smaller openings, helping to limit inflow of fine materials (Royster, 1980). Nevertheless, clogging through infilling of soil particles, encrusting of calcium carbonate and other precipitated minerals, and root growth remains a problem requiring regular maintenance. Drains made of either steel or PVC require a maintenance program to prevent clogging from drastically shortening their life span. A proper maintenance program includes annual inspections of drain outlets, repairs when determined necessary during inspections, and cleaning of the drain pipe every 2 to 8 years, depending on soil grain size and water chemistry (FHWA, 1980). More recently, Cornforth (2005) recommended drain cleaning and inspection every 3 months during the first year, once the following year, and every 4 years thereafter, except where calcium carbonate buildup is expected, in which case the inspections and cleaning should be done every 2 years. Inspections of horizontal drains serve two purposes: The first purpose is to provide an opportunity to identify and correct any issues that may negatively affect drain performance, and the second is to keep the locations of drains from being buried or forgotten as personnel change. Inspections may indicate needed repairs to drain outlets, especially along roadways were they can be damaged by vehicles. Not included in the outlet repair, drain cleaning removes material that limits drain performance, such as root growth, in the first 3 to 5 m (10 to 15 ft) of the drain. This blockage can usually be removed by a cutting blade on the end of a short steel rod (FHWA, 1980). The drain pipes can also become clogged with fine-grained sediment that passes through the perforations or slots but is not carried out of the pipe. In addition, the perforations or slots can become clogged with sediment or precipitated minerals. These sediments and precipitated minerals, both inside the drain and in the slots or perforations, can be removed with a high-pressure jet system such as those designed for clearing sewer lines and culverts (FHWA, 1980). This maintenance, required for traditional horizontal drains, adds significantly to the cost over the lifetime of the project. Additionally, maintenance is often not performed. A new method of horizontal drainage was developed by Santi (1999) and Santi et al. (2001a, 2001b, 2003) to stabilize slopes of fine-grained materials using wick drains in place of PVC pipe. Wick drains are pre-fabricated, ribbon-shaped drains made of synthetic materials that were traditionally used as vertical drains to speed consolidation in low-permeability materials. Santi (1999) and Santi et al. (2001a, 2001b, 2003) adapted the installation method of these drains to be used horizontally in unstable slopes. Unlike pipe drains, wick drains do not require drilling; instead, they are pushed into the slope with the bucket of an excavator or blade of a dozer. This method of installation limits horizontal wick drains to lengths of 45 to 60 m (150 to 200 ft), but it is faster and more economical than PVC pipe drains, and, once installed, the drains do not require maintenance (Santi et al., 2001a, 2001b). Additionally, wick drains elongate by as much as 60 percent before rupture (Nilex Corporation, 2009), allowing the drains to remain functional if movement occurs within the slope. Due to its brittle nature, PVC pipe is likely to break if movement continues after installation, which may prevent proper drainage. The speed of installation and the elongation capabilities make horizontal wick drains ideal for landslide mitigation. Since horizontal wick drains were first developed, the installation procedure has been improved (Santi et al., 2001a, 2001b), the effectiveness at increasing the stability of slopes has been shown (Santi et al., 2003), and drain layout design has been optimized (Crenshaw and Santi, 2004; Cook, 2009), yet the adoption of this drainage method has been slow. This is in part due to an incomplete understanding of the life span of horizontal wick drains. The earliest installed wick drains are now 11 years old, and a better understanding of the effects of clogging is necessary in order to establish whether re-installation is necessary and, if so, under what time frame. BACKGROUND With the development of synthetic materials such as polyester, polypropylene, and polyethylene, a large variety of geotextiles has been manufactured for use in engineered earth projects. Geotextiles are used for drainage, soil reinforcement, and ventilation, as well as other purposes. Wick drains are one geotextile 104 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

3 Life Span of Horizontal Wick Drains product and can be made of a variety of materials, but those studied during this project were made of polypropylene for both the filter and the core of the drain. The filter is a fabric of non-woven fibers, and the core is an extruded sheet corrugated to create drainage channels. The life span of wick drains and other geotextiles is, in part, determined by the life span of their constituent materials, which are affected by ultraviolet light, extreme temperatures, radioactivity, oxidation, and polluted atmospheres (Rollin and Lombard, 1988). Because wick drains are buried in the soil and the outlets are typically protected with exterior-grade plastic pipe, exposure to sunlight is not a concern for long-term durability. The temperatures at Earth s surface are well within the acceptable range of geotextiles (Rollin and Lombard, 1988), and radioactivity is not likely to be a common problem in the natural and engineered slopes where wick drains are used for landslide drainage. Similarly, chemical degradation of wick drains is not expected when used for landslide remediation because the groundwater likely does not contain aggressive chemicals and because polypropylene is considered inert to the soil environment and is stable over a wide range of ph values (Koerner et al., 1988). Because of the durability of synthetic materials, wick drains are expected to persist much longer than the design life of many typical engineered structures. Though the drain materials are not expected to degrade, the functionality of wick drains does decrease once installed. The performance of vertical wick drains has been studied by several researchers to identify potentially adverse factors and to optimize drain and drain system design to speed consolidation. The results of some of these studies are discussed next and are expected to apply as well to horizontal wick drains. One factor that has been found to negatively affect drain performance is the creation of a smear zone, resulting from the installation procedure of driving the drains into the ground with a mandrel. The resulting smear zone is the region of disturbed soil surrounding the drain that has a reduced horizontal hydraulic conductivity due to the realignment and compaction of clay particles. The cross-sectional dimensions of wick drains are typically 4 mm by 100 mm (0.2 in. by 4.0 in.), and the resulting smear zone is approximately 10 mm (0.4 in.) thick (Atkinson and Eldred, 1981). Casagrande and Poulos (1969) concluded that the smear zone in sensitive clays could be large enough to render the drains ineffective. Over time, however, the filter fabric surrounding a wick drain permits clay particles to pass through while retaining the coarser materials. This removes the clay particles from the smeared zone, forming a natural graded filter approximately equal in size to the smear zone and thus restoring the hydraulic conductivity (Atkinson and Eldred, 1981). Clogging also leads to a loss in drain function. Soil particles clog geotextile fabrics by blinding and impregnation (Gosh and Yasuhara, 2004). Blinding occurs when soil particles that are too large to pass through the filter accumulate on the outside of the filter, restricting the flow paths for water to enter the drain. Slightly smaller particles may enter the filter and become lodged within the fabric, known as impregnation. Though impregnation does reduce possible flow paths for water, it can also help maintain transmissivity and in-plane flow in thicker geotextiles that are subjected to confining loads by limiting their compressibility (Palmeira and Gardoni, 2000). Wick drain filters, however, are not expected to compress significantly because they are approximately half a millimeter thick (Nilex Corporation, 2009), and impregnation is thus likely to only decrease wick drain function. Clogging can be caused by other materials as well. The precipitation of iron oxide onto the fibers of the filter can, with time, decrease the water flow through the drain. This oxidation reaction is promoted by the various soil bacteria that use this reaction as part of their life process (Rollin and Lombard, 1988). Salts, various carbonates, and sulfates may also precipitate onto the drain materials. Mineral buildup is most prominent where local groundwater contains significant levels of these minerals and can evaporate upon reaching the drains (Rollin and Lombard, 1988). Long dry periods may exacerbate this problem by allowing crystal growth to continue for extended lengths of time. The effects of clogging on the flow rate through wick drains were examined by Miura and Chai (2000), who looked at the changes in discharge capacity of wick drains confined in clay over a 6 month period. They found a significant decrease due to two factors: creep of the filter fabric into the drainage channels of the filter, reducing the space to carry water, and migration of small soil particles into the filter and drainage channels. Miura and Chai (2000) discovered that the cross-sectional area of a wick drain is reduced by approximately 20 to 60 percent under confinement pressures of 49 to 400 kpa (7 to 58 psi). This pressure range is approximately equivalent to the horizontal soil pressures at 3 to 25 m (10 to 80 ft) depth. Despite this decrease in cross-sectional area, the authors concluded that particle migration was the more significant cause of reduction in discharge by showing that discharge rates could be restored to near initial levels by creating a surge in flow. At the end of the 6 month test, several drains were subjected to a rapid increase in head, creating a surge, which dislodged some particles and restored the discharge to values Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

4 Mininger, Santi, and Short Figure 1. Graph displaying predicted change in discharge of wick drains based on the research of others (from Santi et al., 2001b). 1 cm/s 5 2,835 ft/d. roughly equivalent to those measured after 1 week of confinement. Miura and Chai (2000) also showed that with increasing difference in head across the drain, the reduction in discharge over time was reduced. Thus, in slopes prone to fluctuations in groundwater level, the changes in gradient across the drain may help reduce clogging. One method of reducing the degree of potential clogging is to select the proper drain filter. Wick drain manufacturers provide a choice of filter materials with various-sized openings. The proper relationship between soil grain size and filter opening size is desired, so that the finest particles can pass through the filter and out of the drain, while the larger particles are retained to form a natural soil filter around the drain and prevent piping. Chen and Chen (1986) proposed the following two criteria for selecting the proper filter material, O 90 d 85 v1:2*1:8 O 50 d 50 v10*12 ð1þ ð2þ where O n refers to the opening size in the filter material, and d n is the particle diameter of the soil as calculated by a grain-size distribution analysis. Equation 1 is a permeability criterion that provides a filter with sufficient flow for a given soil, while Eq. 2 prevents filter clogging and promotes the development of the natural soil filter by limiting the size of soil particles that are able to enter the drain while the soil filter is developing around the drain. The Nilex Corporation (Goughnour, 2001) generally recommends only the following relationship, similar to Eq. 1. O 95 ƒ1:8 ð3þ d 85 This is the preferred equation from the wick drain manufacturer because it relies on the 95th percentile opening size, which is defined by ASTM standard 4751 to be the apparent opening size (AOS) of the filter material. This value is calculated and reported for all manufactured wick drain filters, making Eq. 3 a more convenient criterion. When proposing the new use for wick drains as horizontal drains, Santi et al. (2001b) also investigated the effects of clogging. The results of research on the effects of soil smear, particle migration, and confining pressure were combined to produce a graph of projected clogging (Figure 1). This graph shows the trend of initial increase in discharge with time as the effects of smear on the surrounding soil are reduced and then a slow decrease in discharge as clogging occurs. The hydraulic conductivity in the vast majority of natural soils will not be below cm/s ( ft/d), so the reduction in flow should be on the order of 25 percent (approximately 75 percent of maximum flow) after 12 years (Santi et al., 2001b). To investigate clogging further and to attempt to quantify clogging of horizontal wick drains, Santi and Elifrits (2002) developed a laboratory test to simulate in-slope conditions. For 26 months, drains were surrounded by clay soil in a Plexiglas tank while water circulated through under a constant head. Visual examination of these drains after they were 106 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

5 Location Table 1. A summary of the sample locations and soil types according to the Unified Soil Classification System (USCS). Sample Labels Life Span of Horizontal Wick Drains Number of Samples Drain Type Site Description USCS Soil Type Years in Use Blackhawk, CA BH 4 MD-7407 Test slope SC: clayey sand (fill) 1.2 Meeker, CO MCO 3 MD-7407 Road embankment CL: lean clay 10.0 St. Joseph, MO SJMO 2 MD-7407 Road cut CL: lean clay (loess) 10.0 Boonville, MO BMO 1 MD-7407 Road embankment CL: lean clay 10.5 Rolla, MO RMO 2 AMERDRAIN Test slope CL-ML: sandy silty clay (fill) 10.8 Fox Point, WI FPWI 2 MD-88 Lake bluff CL: lean clay (glacial till) 5.3 Bender Park, WI BPWI 2 MD-88 Lake bluff CL: lean clay (glacial till) 4.8 Golden, CO GCO 1 MD Uncompacted fill CL: lean clay 0.3 exhumed revealed that although the exterior of the drains showed packed soil particles, very few particles were visible on the inside of the filter material, suggesting that finer particles were not able to migrate to the inside of the filter and reduce water flow. Santi and Elifrits (2002) also exhumed drain samples after 2 years of use in a lean clay slope near Meeker, CO. Visual examination of these samples revealed fine particles coating the fabric strands on both the inside and outside of the filter. However, when the samples were collected, the drains were producing a continuous flow of water, indicating that these fine particles did not significantly clog the drain or prevent transmission of water. The presence of soil particles on the inside of the drains collected in the field but not in those from the laboratory study may be due to the continuity of drainage. The drains in the laboratory test drained continuously at a constant flow rate. Those in the field would have drained frequently if not continuously but at a fluctuating rate. As will be explained later, drains that flow frequently clog more quickly than those that flow infrequently; however, a constant flow rate, as in the laboratory test, will not allow fine grain particles to settle onto the inside of the filter. Periods of very low flow in the field allow fine particles to adhere to the inside of the drain before being washed from the drain; thus, the drain samples collected from the field showed soil particles on the inside of the drain, while those from the laboratory test did not. It is clear from the work of Santi and Elifrits (2002) that horizontal wick drains can be expected to last more than 2 years, but when used to stabilize landslides or slopes where movement is expected, they are intended for much longer life spans. Figure 1 suggests a life in excess of 12 years, but a more quantitative study is clearly needed. The earliest installed horizontal wick drains have now been in use for 11 years, providing an opportunity to investigate the long-term effect of clogging and to calculate when or if replacement will be necessary in order to prevent failure of the drains. METHODS As the filter material of a wick drain clogs, with transported soil particles, precipitated minerals, or biological growth, the amount of water that can flow through the material decreases. If the flow decreases sufficiently, the drain may no longer adequately lower the groundwater level in a slope and may possibly lead to instability. Because horizontal wick drains cannot be cleaned, they must be replaced before they clog to such a degree. We pursued the following seven steps in order to gain an improved understanding of the clogging of horizontal wick drains and predict their life span. 1. Exhume drain samples from locations with varying soil types and where the drains have been in use for varying lengths of time. 2. Characterize soils from locations where drain samples are collected. 3. Measure the degree of clogging of the drain samples in two ways: N by measuring the increase in drain sample mass over that of a clean sample, and N by estimating the percentage of the filter that is clogged while examining the sample under magnification. 4. Measure the flow rate of drain samples using a constant head test and calculate the reduction in flow rate by comparing the flow rates of exhumed drains to clean drains. 5. Relate the measured degree of clogging to the reduction in flow rate through the drains. 6. Correlate the degree of clogging with the length of time the drain has been in use and to specific soil properties to calculate a clogging rate for various soil types. 7. Use the clogging rate to predict the usable life span of horizontal wick drains. Methods used for these steps are described in more detail next. Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

6 Mininger, Santi, and Short Table 2. Product specifications for the three drain types used in this study. Drain Type Mebra-Drain MD-7407 Mebra-Drain MD-88 AMERDRAIN-407 Manufacturer Nilex Corporation Nilex Corporation American Wick Drain Corporation Drain core material Polypropylene Polypropylene Polypropylene Filter material Polypropylene Polypropylene Polypropylene Drain width (mm) (in.) Core thickness (mm) * (in.) * Composite thickness (mm) * (in.) * Filter elongation (percent) Apparent opening size (mm) (sieve #) Permittivity (s 21 ) *Dimension not reported on product specification sheet but similar to MD Exhuming Drain Samples Seventeen drain samples were collected from eight different locations where they had been in use from 4 months to 11 years (Table 1). These sites included road cuts and fills, natural slopes, and test slopes. The drains at the eight sites ranged in length from approximately 5.5 to 24.5 m (18 to 80 ft). Ideally, drain samples would be collected deep in the slope along the failure plane in slopes where movement has initiated or near the probable zone of failure in slopes where instability is suspected. However, because the slopes could not be destabilized during sampling, only a short section of drain near the outlet could be uncovered in most slopes. This was accomplished by using a shovel to exhume the first meter of the drain, taking care not to smear the drain with the surrounding soil. A 90 cm (3 ft) sample was cut from the drain, replaced with new drain material, and then reburied, leaving the drains functional after sampling. The 90 cm (3 ft) samples represented between 4 and 17 percent of the length of drains from which they were collected. At the test slope in Blackhawk, CA, the procedure was different, and the drains were completely exhumed. In this case, the majority of the soil from the slope was removed with a trackhoe, leaving the last few centimeters (1 to 2 in.) to be removed by hand with a shovel. Two samples, approximately 90 cm (3 ft) in length, were collected from each of the two drains in the slope, one from the base and the other near the top of the slope. The samples collected from the base of the slope, near the outlet, showed the highest degree of clogging for this slope. This demonstrated that samples collected from close to the outlet are representative of the maximum degree of clogging within the slope. Three different wick drain types were sampled during this project, two manufactured by the Nilex Corporation and one by the American Wick Drain Company. Relevant manufacturer specifications for the three drain types are provided in Table 2. The majority of the drains were Nilex drain style MD-7407, those exhumed in Wisconsin were Nilex MD-88, and those in Rolla, Missouri were AMERDRAIN 407 by the American Wick Drain Company. All three drain types function similarly but vary in the pore size of their filters and the design of the drain core, changing the size and shape of the drain channels (Figure 2). Most significant to this study, the filters of the three drain types vary in their apparent opening size (AOS). The AMERDRAIN had an AOS equivalent to a #80 sieve (0.18 mm), MD-7407 was equivalent to a #110 sieve (0.14 mm), and the MD-88 drain filters were equivalent to a #170 sieve (0.09 mm). Horizontal drains do not typically carry water continuously in the field, often producing water only during the wet seasons and following precipitation events. The drains, filter material, and any clogging soil particles have thus likely gone through multiple drying and wetting cycles. As a result, no attempt was made to preserve the in situ moisture content of the drain samples, and they were allowed to dry before testing. They were then cut into three pieces: 35 cm (14 in.) for the flow test, 10 cm (4 in.) for microscope inspection, and 40 cm (16 in.) for mass measurements. Soil Classification In order to characterize the soil surrounding each drain sample, a sample was collected from each site where drains were unearthed. A grain-size distribution 108 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

7 Life Span of Horizontal Wick Drains Figure 2. Photos of two drain types sampled, MD-7407 (left) and MD-88 (right), both manufactured by the Nilex Corporation. The apparent opening size (AOS) of the filter around the MD-7407 drain is equivalent to a #110 sieve (0.14 mm), and the AOS of MD-88 was equivalent to a #170 sieve (0.09 mm). A third drain type, AMERDRAIN TM 407, was also sampled and looks similar to an MD-7407 (left) but has a filter AOS equivalent to a #80 sieve (0.18 mm). analysis was conducted, and the Atterberg limits were measured for each soil sample in general accordance with ASTM standards D 422 and D 4318, respectively. Results were used to classify the soils according to the Unified Soil Classification System (USCS). The grainsize distributions were also used to calculate the 50th and 85th percentile soil particle diameters, d 50 and d 85, which were compared with the filter opening sizes using the suggested criteria in Eq. 1 through Eq. 3. Soil analyses were performed by the authors, except for those on the soil collected at Blackhawk, CA, which were performed by Kleinfelder, Inc. Tests performed on the Blackhawk soil included hydraulic conductivity analyses as well. Degree of Clogging by Drain Mass The first method of assessing the degree of clogging of the drain samples was to measure their mass. This measurement could then be compared to the mass of a length of clean drain to quantify the amount of material clogging the drain. To ensure that the measured increase in mass was only due to material that was clogging the drain, the samples were prepared by first removing the roots and loose clumps of soils on the outside of the filter material by hand. This was done by identifying and removing any material that was physically separated from the filter and any thick layers of soil buildup. The layer of soil immediately next to, adhered to, or impregnating the filter was not removed. Finally, the drain was brushed with a soft bristled laboratory brush to remove any remaining loose material. While cleaning the outside of the filter, care was also taken not to dislodge any material from inside the drain. Degree of Clogging by Inspection under Magnification The second assessment of clogging was made by magnified inspection of both the inside and outside of the filter material. A Leitz Laborlux 12 POL petrographic microscope with 353 magnification was used to examine the filter, while digital images were captured with an attached Leica EC3 digital camera and processed using Leica s LAS EZ version software. A 3 by 3 grid pattern at 1.25 cm (0.5 in.) spacing was laid out on both the outside and inside of the filter material of each drain sample. Points on the grid were then examined, and the percent of area clogged was estimated using a percent abundance chart typically used for estimating percentages of minerals in rock samples. The average percent of area clogged was then calculated for both the outside and inside of the filter by adding up the percent clogged value from each point on the grid and then dividing by the number of points observed. Constant Head Test A constant head test apparatus was developed to measure the flow rate through wick drain samples (Figure 3). A hole with a slightly larger diameter than a rolled drain was drilled into the side of a five gallon bucket. A drain sample was inserted into the hole and then the hole was sealed with caulk to prevent Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

8 Mininger, Santi, and Short Figure 3. A schematic and photograph of the constant head test apparatus. The rolled and tied drain sample (Figure 4) is completely covered in sand to better simulate buried field conditions. The water level is maintained at 10 cm (4 in.) above the drain by the overflow port. leakage. A second smaller hole was drilled 10 cm (4 in.) above the drain to act as an overflow. A second bucket, with a valve to control flow, was placed above the first bucket and used in combination with the overflow to maintain a constant level of water in the first bucket. All water used was stored at ambient air pressure and temperature for a minimum of 24 hours to allow entrained air to escape prior to use. Water draining through either the overflow or through the drain sample was returned to the upper bucket to minimize the amount of water used in the experiment. The bucket with the drain sample was filled with coarse sand to a height of 2 cm (0.8 in.) above the drain sample in order to prevent soil particles on the sample from being washed off and to more accurately replicate field conditions where the full length of the drain would be in contact with the surrounding soil. Drain samples were prepared for the constant head test by rolling and tying them with cable ties, as they would be when used in the field (Figure 4). One end was then sealed using Plastic Dip TM to prevent water from entering. This plastic sealer was also used to seal the section of the filter where the drain passed through the wall of the bucket. The distance between the two sealed areas of the drain was measured for each drain sample and recorded as the wetted length, the length of drain in contact with the water. During the test, the volume of the water that passed through the drain was recorded along with the length of time over which the volume was measured. The final measurement was the duration of time from the beginning of the test until the volume measurement was taken. These measurements were recorded until the flow stabilized, typically requiring between 40 and 80 minutes. The flow rate was calculated by dividing the measured volume of water by the time during which the volume was collected. The flow rate was then normalized to the wetted length and termed the normalized flow rate, so that values from different length samples could be compared. The flow rate through a clean drain was measured for each drain type so that the reduction in flow rate could be calculated for each sample exhumed from the field. RESULTS AND INTERPRETATION While following the previously described steps, observations were made of the clogging materials as 110 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

9 Life Span of Horizontal Wick Drains Figure 4. A wick drain sample prepared for the constant head test. The cable ties maintain the drain in its rolled configuration as it is in field installations. Plastic Dip TM was used to seal one end of the drain, as well as a second area where the drain passes through the bucket. White caulk visible on the sealed area was used to seal the hole around the drain when installed in the bucket. The wetted length is the length of the drain inside the bucket through which water enters. well as the factors affecting clogging. These observations will be discussed in more detail next. A summary of the results, including both measurements of the degree of clogging and the measured reduction in flow rate from the constant head test, is listed in Table 3. Precipitant and Particle Buildup The mineral buildup that is often found in drain pipes as a result of precipitation was not identified in the examined wick drain samples. In many drain fields, plastic pipe is used as a sheath to protect horizontal wick drain outlets from root growth and from exposure at the surface. At the Meeker, CO, site, the outlet pipe contained significant calcium carbonate deposits, reducing the cross-sectional area of the open drain pipe by nearly half. Interestingly, the wick drain samples from this same location did not show any signs of clogging from precipitants. Neither calcium carbonate nor iron oxide made up a significant portion of the particles seen on the filter material (Figure 5). Wick drains appear to resist this form of clogging, Table 3. A summary of the test results and observations. The combined value of the clogged filter was calculated by averaging the inside and outside values but weighting the inside value twice, as explained in the text. 1 g/cm lb/in. Sample Time in Use (years) Increase in Drain Mass (g/cm) Inside (percent) Clogged Surface Area of Filter Outside (percent) Combined (percent) Reduction in Flow Rate (percent) Fine Grain Fraction (percent) Drainage Frequency BH Frequent BH Intermittent BH Frequent BH Intermittent MC Frequent MC Frequent MC Intermittent SJMO Frequent SJMO Intermittent BMO Intermittent RMO Intermittent RMO Frequent FPWI Frequent FPWI Frequent BPWI Frequent BPWI Frequent GCO Intermittent Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

10 Mininger, Santi, and Short Figure 5. Magnified (353) images of outside (left) and inside (right) of wick drain filter showing 90 percent of the surface area clogged with silt- and clay-sized particles, Meeker, CO. This drain was exhumed after 10 years in the ground. Even for drains with this degree of clogging, substantial flow capacity remains (Figures 6 and 7). perhaps due to small drainage channels that do not transport large volumes of air. The environment remains at nearly 100 percent relative humidity, preventing the evaporation of water and precipitation of dissolved minerals. Soluble minerals thus remain in solution and are carried out of the wick drain, depositing in the outlet pipe or on the ground surface. The principal materials responsible for clogging the drain samples were fine sand-, silt-, and clay-sized soil particles. Buildup was heaviest on the exterior of the filter fabric, although most drains showed noticeable quantities of soil particles on the interior of the drains as well (Figure 5 and Table 3). As mentioned before, an average percent of area clogged was calculated for both the outside and inside of the filter. After comparing both clogging values with the measured reduction in flow rate, we observed that a combination of the degree of clogging on the inside and the outside of the filter was a better indicator of the measured reduction in flow rate than one measurement alone. The best correlation between area clogged and the measured reduction in flow rate was found by weighting the values from the inside of the drain by a factor of two. This was done by averaging the percent clogged values from the outside and the inside but counting the inside values twice, as in the following equation: DC visual ~ PC outsidez2(pc inside ) 3 ð4þ where DC visual is the degree of clogging estimated from visual examination, and PC outside and PC inside are the average percent of filter area clogged on the outside and inside, respectively. As would be expected, clogging material decreases the rate of water passing through the filter and out of the drain. The measurements of clogging, both by increased drain mass and by the estimate based on visual appearance, have a positive relationship with the reduction in flow rate (Figure 6 from data in Table 3). Both relationships display an inverted decay curve, demonstrating the expected diminishing effect of increased clogging material. The filter fabric is made up of non-woven fibers that provide numerous flow paths. The first soil particles to reach the fabric block the most direct paths and have the greatest effect on reducing the flow rate. As more material reaches the filter, it accumulates but has less effect than the initial particles. An upper and lower bound represented by the 85th percentile prediction interval (Figure 6) is the range of expected reduction in normalized flow rate given the degree of clogging, measured either visually or by increase in drain mass. To calculate a reduction in drain flow via a simpler process than the timeconsuming and cumbersome constant head test, Figure 6 could be used to select the corresponding value from either of the measures of degree of clogging. Both methods appear to be equally accurate, with comparable r 2 values and comparable average values. To be conservative, the reduction in flow value along the 85th percentile upper boundary should be chosen to represent the maximum reasonable expected clogging. Factors Influencing Clogging Clogging, measured as the reduction in the flow rate through the drain, was calculated for each drain 112 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

11 Life Span of Horizontal Wick Drains Figure 6. The relationships between the reduction in flow rate per length of drain and the two measures of the degree of clogging: the increase in drain mass (left), and visual estimation under magnification (right). Upper and lower bounds represent the prediction interval with 85 percent confidence. Both methods of quantifying the degree of clogging appear equally accurate. 1 g/cm lb/in. sample by comparing the measured flow rate of that sample and the flow rate through a clean sample of that same drain type. The flow rate per length of clean drain for both the MD-7407 and the AMERDRAIN 407 was 1.7 ml/s?cm ( gal/s?in.), and for drain type MD-88 it was 1.6 ml/s?cm ( gal/s?in.), with an average of 1.65 ml/s?cm ( gal/s?in.). The decreased drain flow of the MD-88 drains is likely due to a combination of smaller openings in the filter and slightly smaller flow channels in the drain core. Several factors affect the amount of clogging that drains experience. Primarily, the longer a drain is in use, the higher is the degree of clogging. The relationship between clogging and time of use is not linear, however, and the majority of clogging occurs early in the life of the drain. The clogging rate decreases with time, and correspondingly the reduction in flow rate slows and levels off once further clogging has ceased (Figure 7a from data in Table 3). A decreasing rate of clogging over time is expected in a soil environment where a natural filter develops around the drain. Once this filter has developed, very few new soil particles reach the drain filter. Furthermore, the decreased water flow from initial clogging carries fewer particles to the drain, limiting continued clogging. More frequent drainage also leads to more clogging and a corresponding reduction in the normalized flow rate (Figure 7b). Categorizing the drain samples into either frequent drainage or intermittent drainage produces two distinct groupings. The label of frequently draining was applied to drains that, when exhumed, were actively draining, the soil around them was saturated, or there was algal growth at the outlet, indicating constant moisture input. Those drains in moist or dry soil or those above a lower row of drains were labeled as intermittently draining. Drain samples with intermittent drainage had reductions in normalized flow rate of 11 percent or less, while frequently draining samples had reductions ranging from 35 to 85 percent. This distinction was not correlated with time of use, because the full range of 1 to 11 years of use was represented in both categories of drainage frequency. The observed increase in clogging with frequent drainage is the result of more groundwater being able to transport more soil particles. Presumably, drains with intermittent flow will reach the same degree of clogging as drains with frequent flow, only over a longer time period. Soil characteristics also influence the maximum degree of clogging. The percentage of fine particles, those finer than mm (#200 sieve), relates significantly with the measured reduction in flow rate through the drain samples (Figure 7c). Drains in the lean clay soils, with 85 percent or more fines, had much more reduced flow rates than the soils with 35 to 60 percent fines. The lowest normalized flow rate measured for drains in soils with 85 percent or higher fines was 0.26 ml/s?cm ( gal/s?in.), a reduction in flow rate of nearly 85 percent from that of a clean drain. The lowest normalized flow rate for drains in soils with 35 to 60 percent fines was 0.99 ml/s?cm ( gal/s?in.), a reduction of 42 percent. One drain was sampled from soil with 77 percent fines content, but this drain exhibited zero reduction in flow rate due to very infrequent flow. With no Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

12 Mininger, Santi, and Short Figure 7. The clogging rate of horizontal wick drains decreases with time, as demonstrated by a leveling-out of the reduction in flow rate. (a) The relationship between the reduction in flow rate per drain length and the time of use of the drains, with the upper trend leveling off as it approaches 85 percent reduction. (b) Drains experiencing more frequent flow (defined in the text) have the greatest clogging. (c) Plot showing the higher clogging potential of drains in predominantly fine-grained soils. It was unclear which category the drain with 77 percent fines would fall into, since it did not drain frequently and so experienced very little clogging. (d) Same as Figure 7a with the data grouped by drainage frequency and fines content. Three ranges of reductions in flow rate are apparent. reduction in flow rate measured in the drain, it was unclear whether a soil with 77 percent fines should be classified as moderate fines content or high fines content. This sample was therefore not grouped with other samples (Figure 7c). Finer soils cause more reduction in flow because the smaller particles are able to impregnate and blind the filter more efficiently. Palmeira and Gardoni (2000) suggest that gap-graded coarse soils with significant percentages of fines may also have high clogging potential because they are not able to form natural graded filters. No soils that fit this description were sampled during this project, and the distribution of such soils is expected to be limited. In addition, the hydraulic conductivity of coarse soils would likely be high, and thus horizontal wick drains would not be the preferred drainage choice. For these reasons, soils with less than 35 percent fines are not expected to clog and reduce the normalized flow of horizontal wick drains more than the soils that were tested in this study. Grouping the data by the clogging potential and draining frequency helps to explain the large spread of reduction in flow rate values in Figure 7a. Drains with frequent drainage follow two trends (Figure 7d); those in soils with 85 percent or more fine grains have a reduction in normalized flow rate of 45 to 85 percent, and those in soils with moderate amounts of fine particles have reductions of 35 to 42 percent. Drains with intermittent drainage have reductions of 0 to 11 percent, regardless of fines content in the surrounding soils. These drains do not demonstrate the maximum expected clogging and were not used for predicting future clogging behavior of drains, as discussed in the next section. The compliance of the sampled drain filters with the filter criteria proposed by Chen and Chen (1986) was examined using filter opening size distributions derived from Chen and Chen s (1986) study. From the 90th and 50th percentile pore sizes for each filter, Eq. 1 and Eq. 2 were used to calculate the smallest 114 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

13 Life Span of Horizontal Wick Drains Figure 8. (a) The filter opening size distribution curves (based on filter analysis by Chen and Chen, 1986) of the three drain types sampled (curved lines), along with points indicating the smallest allowable grain size according to the filter criteria by Chen and Chen (1986), superimposed on the soil particle size distribution curve of the soil from Blackhawk, CA. Soils from Meeker, CO (b), and St. Joseph, MO (c), were chosen to demonstrate the effect of the criteria on long-term clogging because both of these soils had high fine particle content, and drains in these soils were the same type, were comparable in their frequency of drainage, and were in use for the same length of time. The grain-size distribution of the soil from Meeker, CO, is close to the limit set by the filter criteria, and drains in this soil experienced up to an 85 percent reduction in flow rate. The soil from St. Joseph, MO (c), does not approach the limit for the 50th percentile soil particle, and drains in this soil had maximum reductions in flow rate of 61 percent. 1 mm in. 85th and 50th percentile soil grain size that would meet the criteria (Figure 8). Generally, those soils with grain-size distributions that plot closest to the limits set by the criteria had the highest degrees of clogging. Two of the soil/filter combinations did not fit the criterion set out in Eq. 1, i.e., a ratio of 90th percentile opening size (O 90 ) to 85th percentile soil grain size (d 85 ), 1.8. Both of the soils were close, however, with ratios of 1.9 and 2.1, and neither experienced excessive clogging. All soils met the criterion of Eq. 2, except the sample collected from Golden, CO, where the maximum degree of clogging was not observed because of infrequent drainage and a short time in use. These results support the filter criteria proposed by Chen and Chen (1986), especially Eq. 2, which limits the size of the 50th percentile filter openings in relation to the 50th percentile soil particle diameter. All drains that experienced frequent flow and thus a higher degree of clogging met this criterion and did not clog to the point of failure, as evidenced by drain flow at the time of exhumation. ANALYSIS AND DISCUSSION The reduction in normalized flow rates through drains in soils with less than 60 percent fine particles forms an asymptotic curve that approaches 42 percent (Figure 7d). This curve is initially very steep, with nearly all the clogging occurring during the first year. It then levels off, with very little additional clogging over time. Soils with high fines content also display an asymptotic curve, with the maximum reduction in flow rate approaching 85 percent. This curve again begins steep but transitions more gradually, with only about half of the clogging occurring during the first year. An 85 percent loss in drain function is significant, but the important question is whether the remaining 15 percent of original flow is sufficient to drain an unstable slope of soil with high fines content. Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

14 Mininger, Santi, and Short Figure 9. General water-table profiles along a horizontal drain, where v is groundwater recharge rate, and K is the hydraulic conductivity of the soil. Type II represents the expected range in the field (modified from Crenshaw and Santi, 2004). In order to address the question of the degree of clogging that will cause drain failure, the flow rate of the drain must be related to the amount of groundwater needed to be drained from the slope. Crenshaw and Santi (2004) examined this relationship while defining the shape of the groundwater surface in a drain field. They created laboratory models of drained slopes with various soil types and under various groundwater recharge scenarios. The results of these studies defined three general groundwater profiles (Figure 9). Type I is a low-groundwater-flow scenario, where the groundwater flow rate is two orders of magnitude lower than the hydraulic conductivity of the soil or lower. A type II profile has a ratio of groundwater recharge rates to hydraulic conductivity between 0.01 and 0.30, and it is considered to be the most representative of expected field conditions (Crenshaw and Santi, 2004). Type III profiles occur under conditions of very high groundwater flow rate, with a ratio of 0.30 or higher. In high-groundwater-flow conditions, perforated PVC drains would be a more appropriate drainage system than horizontal wick drains because they can carry much larger volumes of water. A ratio of groundwater flow rate to hydraulic conductivity of 0.30 thus represents the highest expected groundwater flow for horizontal wick drains. Crenshaw and Santi (2004) also created a computer model of the groundwater surface to understand the factors that were affecting its behavior. Drains were modeled as ideal, when their conductance was set such that they were able to carry away all water that reached them without negatively impacting the performance of the drain. These models related drain flow rate, drain spacing, drain length, and groundwater recharge rate with the following equation: v~ Q L:S ð5þ where v is the recharge rate of water entering the drain field in dimensions of length divided by time (L/T); Q is the volumetric flow rate exiting the drain (L 3 /T); L is the drain length (L); and S is the horizontal drain spacing (L). Equation 5 can be rearranged to solve for the flow rate per length of drain or normalized flow rate. Q n ~ Q L ~v :S ð6þ By replacing the groundwater recharge rate (v) with the ratio between groundwater flow rate and hydraulic conductivity of 0.30 discussed previously, Eq. 6 can be rearranged to calculate a critical normalized flow rate, Q n crit : Q n crit ~0:3K:S ð7þ where K is the hydraulic conductivity of the surrounding soil (L/T). The critical normalized flow rate represents the volume of water per unit time that must enter the drain per unit of drain length. If the drain meets or exceeds the critical normalized flow rate, groundwater will be carried away as it is introduced by the soil. If this critical value is not met, water will not be drained quickly enough, and the water table will rise in the slope. A critical normalized flow rate can thus be calculated for any slope given the hydraulic conductivity of the soil and drain spacing. Proper drain spacing was outlined by Cook et al. (2008), who suggested a spacing of 8 to 15 m (26 to 116 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

15 Life Span of Horizontal Wick Drains Figure 10. The critical flow ratio of the normalized flow rate of a drain with maximum clogging to the critical normalized flow rate found from a given drain spacing and hydraulic conductivity. Hydraulic conductivity values shown in the chart are in cm/s (1 cm/s 5 2,835 ft/d). The critical flow ratio is above two for all expected conditions, demonstrating that even with maximum clogging, the drains will be able to carry greater than two times more water than necessary to maintain the designed reduction in groundwater table. Line color indicates the maximum clogging used for the different hydraulic conductivity soils based on expected fines content. 1 m ft. 49 ft) for high-permeability materials, such as those with moderate fines content, and a reduced spacing of 1 to 8 m (3 to 26 ft) for soils with low permeability and high fines content. The maximum flow requirement through a drain would result from using the fewest drains, a spacing of 15 m (49 ft), in a soil with the highest expected hydraulic conductivity. Of the soil samples collected during this project, the hydraulic conductivity was measured only for the clayey sand soil from Blackhawk and was found to be cm/s ( ft/d). This soil was the coarsest soil studied and thus likely had the highest permeability. It is conceivable that wick drains may be used in soils with higher permeability, such as silty sands with hydraulic conductivities up to of cm/s (2.8 ft/d) (Das, 2008). For a soil with this permeability and drain spacing of 15 m (49 ft), the critical normalized flow rate calculated using Eq. 7 is 0.45 ml/ cm?s ( gal/s?in.). As mentioned already, the lowest normalized flow rate measured for a sandy soil with maximum observed clogging was 0.99 ml/cm?s ( gal/s?in.). This measured flow rate is more than two times the critical value, representing a critical flow ratio, or safety factor, of two. The critical flow ratio is defined as the ratio between the normalized flow rate of a clogged drain and the critical normalized flow rate that the drain must carry. Using the range of expected hydraulic conductivity from to cm/s (2.8 to ft/d), and drain spacing of 1 to 15 m (3 49 ft), we used Eq. 7 to calculate the critical normalized flow rate for expected hydraulic conductivity and drain spacing combinations. The reduction in flow rate of 85 percent for fine-grained soils and 42 percent for sandier soils (Figure 7d) was then multiplied by the normalized flow rate through a clean drain, 1.65 ml/s?cm ( gal/s?in.) to calculate the reduced normalized flow rate through a sample experiencing maximum clogging. The critical flow ratio could then be calculated for the full range of expected hydraulic conductivities (Figure 10). For clay soils with low hydraulic conductivities, to cm/s ( to ft/d), this ratio is several hundred to several thousand. For coarser soils with hydraulic conductivities between and cm/s (2.8 to ft/d), the critical flow ratio ranges from two to several hundred. Thus, under expected conditions, drains that experience the maximum anticipated degree of clogging will continue to be able to carry away the critical amount of water and function properly. Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

16 Table 4. The critical reduction in flow rate is the percent reduction in normalized flow rate necessary to reduce the drain flow to the critical normalized flow rate, which is the rate necessary for intended reduction of the groundwater table. The critical values listed are higher than the measured reductions in flow rate, and therefore drains are unlikely to clog to the degree that they cannot carry sufficient water to maintain functionality. 1 cm/s 5 2,835 ft/d; 1 m ft. Hydraulic Conductivity (cm/s) Mininger, Santi, and Short Drain Spacing (m) * * { { { *Maximum measured reduction in flow rate is 42 percent for soils with moderate fines content. { Maximum measured reduction in flow rate is 85 percent for soils with high fines content. The next question to be answered is If the expected clogging and reduction in flow rate will not negatively affect drain performance, how much reduction in flow rate would be necessary to cause drain failure? The reduction in flow rate that would result in drain failure is termed the critical reduction in flow rate and represents the amount, in percent, by which the clean normalized flow rate must be reduced to equal the critical normalized flow rate. Any clogging beyond this critical value would result in reduced drain function and possible slope instability. To this end, we used the following equation: Critical reduction in flow rate~ 1{ Q n crit 100 ð8þ Q n clean where Q n clean is the normalized flow rate through a clean drain sample. This analysis demonstrates that in the vast majority of situations, the critical reduction in flow rate is near to or above 99 percent (Table 4). Only drains in soils with the highest hydraulic conductivity, cm/s (2.8 ft/d), and drain spacing wider than 5 m would fail, with a reduction in flow rate of less than 90 percent. However, soils with such high hydraulic conductivity would also have a moderate percentage of fine particles and so would not be expected to clog more than 42 percent. The lowest calculated critical degree of clogging for these drains is 73 percent, and thus they are not expected to fail. Regardless of the degree to which a drain is clogged, there is a limit to the volume of water that a drain can evacuate from the slope during a given time. For the drain to function properly, the amount of water introduced to the drain over its full length must not exceed the maximum carrying capacity of the drain. This drain flow capacity can be calculated by treating the wick drain as a rough pipe and solving for turbulent flow through the drain using Eq (White, 2008). where and Q~ n :Re:p:d 4 " p Re~{ ffiffiffiffiffiffi!# e=d 8:j log 3:7 z 1:775 pffiffiffi j ð9þ ð10þ j~ g :d 3 :h f L:n 2 ð11þ In these equations, Q is the drain flow rate; n is the kinematic viscosity of water; Re is the Reynolds number of flow through the drain; d is the drain diameter; j is the head loss along the drain; e/d is the ratio of pipe roughness to diameter; h f is the head loss due to friction; and L is the drain length. The head loss divided by the drain length is the gradient along the drain, and so Eq. 11 can be re-written as: j~ g :d 3 :i n 2 ð12þ where i is the groundwater gradient along the drain. Incorporating this equation into Eq. 9 and Eq. 10, it is clear that the drain flow rate is proportional to the square root of the gradient rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Re~{ 8: g :d 3: i n 2 log e=d 3:7 z r 1:775 6 ffiffiffiffiffiffiffiffiffiffiffic7 4 g:d 3 A5 ð13þ :i n >< rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n { 8: g :d 3 :i n 2 logb e=d 3:7 z r 1:775 >= ffiffiffiffiffiffiffiffiffiffiffic7 4 g:d 3 A5 p:d >: :i >; Q~ n 2 4 ð14þ 118 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

17 Life Span of Horizontal Wick Drains Figure 11. The maximum drain length is affected by the critical normalized flow rate because drains can transport a finite volume of water from the slope in a given time period. The more water that enters per length of drain, the shorter the maximum length may be. Viable drain length and spacing combinations must plot in or below the appropriate soil hydraulic conductivity band, with unit of cm/s (1 cm/s 5 2,835 ft/ d). The bands represent the expected reasonable range of maximum (upper bound) and minimum (lower bound) groundwater gradients discussed in text. For drains in soils with hydraulic conductivities less than cm/s, this chart need not be consulted because maximum drain lengths are longer than is presently possible given current installation methods. 1 m ft. Many of these factors are difficult to ascertain for wick drain installations, so a sample of drain was tested in the flow rate apparatus with both ends open, as a pipe. The flow rate produced from this drain, 89.1 ml/s (0.024 gal/s), was generated under a 10 cm (3.9 in.) head and through 22.3 cm (8.8 in.) of drain, resulting in a gradient of To compare this gradient to that expected in the field, we can return to the groundwater profiles (Figure 9) developed by Crenshaw and Santi (2004). As discussed earlier, type II is the groundwater profile most expected in the field. The groundwater gradients associated with type II profiles range from 0.1 to The measured drain flow rate of 89.1 ml/s (0.024 gal/s) can be calibrated using the corresponding gradient of 0.45 to find the flow rate at the minimum and maximum gradients expected in the field. The resulting range of drain flow rates associated with this range of gradients is between 42.0 and 98.5 ml/s (0.011 gal/s and gal/s) as calculated by a proportion of the square roots of the gradients (as noted by Eq. 14): pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffi i adjusted 0:1 Q 0:1 ~Q measured p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ~89:1 ml=s pffiffiffiffiffiffiffiffiffi i measured 0:45 ~42:0 ml=s ð15þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi i adjusted 0:55 Q 0:55 ~Q measured p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ~89:1 ml=s pffiffiffiffiffiffiffiffiffi i measured 0:45 ~98:5 ml=s ð16þ where Q 0.1 and Q 0.55 are the adjusted flow rates with a gradient of 0.1 and 0.55, respectively; Q measured is the flow rate measured from an open drain; i adjusted is the desired gradient of 0.1 or 0.55; and i measured is the gradient along the drain with measured flow rate. With this range of drain flow rates and the critical normalized flow rate, a maximum drain length can be calculated. If this length is exceeded, the drain will intercept more water than it can reasonably carry. Equation 17 states that the total flow of the drain results from the product of the drain length and normalized flow rate. The maximum drain length is calculated by solving Eq. 17 for L and using the critical normalized flow rate that must enter along each unit length of drain in order to maintain lowered groundwater levels. Q~L:Q n L max ~ Q Q n crit ð17þ ð18þ Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

18 Mininger, Santi, and Short In the previous equation, L max is the maximum drain length. The results of Eq. 18, when both drain flow rate values from Eq. 15 and Eq. 16 are incorporated, can be used to produce a series of graphs where the drain spacing and length can be checked given the hydraulic conductivity of the soil (Figure 11). An acceptable drain spacing and length combination must plot within or below the band representing the hydraulic conductivity of the soil. A combination plotting above the appropriate band would result in the flow capacity of the drain being overwhelmed by the groundwater flow. For hydraulic conductivities of cm/s ( ft/d) or lower, the calculated maximum drain lengths for all drain spacing values are longer than the 45 to 60 m (150 to 200 ft) maximum length permitted by the installation method (Santi et al., 2001b), and Figure 11 need not be consulted. For those drains in soils with conductivity values between and cm/s (2.8 to ft/d), the chart is useful to prevent the drainage capacity of the drains from being exceeded. It should be noted that while Figure 11 represents the maximum spacing to prevent overwhelming the drains when potential clogging is considered, tighter drain spacing may be required in order to adequately lower groundwater to design levels. Drain spacing design criteria can be found elsewhere, such as Cook (2009). It is also clear from Figure 11 that soils with a hydraulic conductivity of cm/s (2.8 ft/d) require such short drain lengths or close drain spacing as to be impractical in most situations. The corresponding band is maintained on this graph, however, for reference when plotting soils with smaller hydraulic conductivities. CONCLUSIONS The adoption of horizontal wick drains to reduce the groundwater level in unstable slopes will benefit from analysis of the life span of this drainage technique. Unable to be cleaned, these drains must be replaced before they fail. This paper examines the clogging of horizontal wick drains and demonstrates that, under the field conditions in which they are expected to be used, failure due to clogging is not anticipated. The life span of horizontal wick drains is not controlled by the durability of the drain materials, since buried or protected drain materials did not appear to degrade during the 11 year span they were in use. The findings of this project suggest that clogging by soil particles or precipitated minerals also will not limit the life span of horizontal wick drains in the majority of field conditions in which they are used. Wick drains resist clogging due to the texture of the filter fabric and the limited space within the drainage channels. The nonwoven fibers of the drain filter create a vast number of flow paths that provide sufficient flow despite some clogging. Precipitants do not appear to deposit onto the drains in significant quantities due to the small channel spaces, which do not permit adequate air circulation to evaporate the drain water and thus precipitate minerals. Measurements on exhumed drains that have been in the ground for up to 11 years show that the majority of the clogging happens soon after they are installed. As the surrounding soil develops a natural filter around the drain, fewer soil particles reach the drain, and the clogging rate slows. This soil filter eventually protects the drain from further clogging. Higher volume and frequency of groundwater flow result in drains clogging faster, but the soil filter around the drain also develops more quickly and the maximum degree of clogging is unchanged. The filter selection criterion proposed by Chen and Chen (1986) in Eq. 2 is supported by the results of the study. Drains with soil/filter combinations that resulted in filter opening to grain-size ratios close to the limits set by the criterion experienced the greatest degree of clogging but continued to function. The type of soil that surrounds the drain affects the soil filter development and thus the maximum degree of clogging of the drain. For soils with 60 percent or less fine particles, the maximum measured reduction in flow rate was 42 percent. For soils with 85 percent or more fines, the maximum reduction in flow rate was 85 percent. In both cases, the remaining drain flow capacity was sufficient to continue draining the slope. Within the expected soil hydraulic conductivity range of to cm/s (2.8 to ft/d) and a drain spacing between 1 and 15 m (3 and 49 ft), the drains are capable of transporting more than twice the critical flow rate needed to maintain the intended groundwater reduction for slope stability, even with the maximum degree of clogging. For most situations, the flow through the drains would need to be reduced by more than 99 percent before drain function would be affected. By treating the horizontal wick drain as a rough pipe with turbulent flow, we were able to calculate the range of expected flow through the drain. This range was then used to calculate the maximum drain length before drain capacity was exceeded, given the hydraulic conductivity of the surrounding soil and the drain spacing. For drains in soils with hydraulic conductivity values between and cm/s (2.8 and ft/d), we developed a chart (Figure 11) to aid the user in selecting the proper drain length and spacing combination so that 120 Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp

19 Life Span of Horizontal Wick Drains the capacity of the drains is not exceeded. This analysis also indicates that horizontal wick drains are impractical for use in soils with a hydraulic conductivity of cm/s (2.8 ft/d). Soils with such high groundwater flow rates require prohibitively short drain lengths and close drain spacing. ACKNOWLEDGMENTS Research and field work in California, Colorado, and Missouri were supported by Blackhawk Geologic Hazard Abatement District. Field work in Wisconsin was supported by the Gillen Company. REFERENCES ATKINSON, M. S. AND ELDRED, P. J. L., 1981, Consolidation of soil using vertical drains: Geotechnique, Vol. 31, No. 1, pp CASAGRANDE, L. AND POULUS, S., 1969, On the effectiveness of sand drains: Canadian Geotechnical Journal, Vol. 6, No. 3, pp CHEN, R. H. AND CHEN, C. N., 1986, Permeability characteristics of prefabricated vertical drains. In Proceedings of the Third International Conference on Geotextiles: Vol. 3, Vienna, Austria, pp COOK, D. I., 2009, Improvements in Horizontal Drain Design: Unpublished Ph.D. Dissertation, Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO, 131 p. COOK, D. I.; SANTI, P. M.; AND HIGGINS, J. D., 2008, Horizontal landslide drain design: State of the art and suggested improvements: Environmental & Engineering Geoscience, Vol. 14, No. 4, pp CORNFORTH, D. H., 2005, Landslides in Practice: Investigation, Analysis, and Remedial/Preventative Options in Soils: John Wiley & Sons, Inc, Hoboken, NJ, 596 p. CRENSHAW, B. A. AND SANTI, P. M., 2004, Water table profiles in the vicinity of horizontal drains: Environmental & Engineering Geoscience, Vol. 10, No. 3, pp DAS, B. M., 2008, Fundamentals of Geotechnical Engineering, 3rd ed.: Thomson Learning, Toronto, Canada, 621 p. FHWA (FEDERAL HIGHWAY ADMINISTRATION), 1980, The Effectiveness of Horizontal Drains: FHWA Report FHWA/CA/ TL-80/16, 71 p. GOSH, C. AND YASUHARA, K., 2004, Clogging and flow characteristics of a geosynthetic drain confined in soils undergoing consolidation: Geosynthetics International, Vol. 11, No. 1, pp GOUGHNOUR, R. R., 2001, PVDrain: A Computer Program for Analysis of Consolidation Settlement of Soft Ground Treated by Prefabricated Vertical Drains: Nilex Corporation, Englewood, CO, 118 p. KOERNER, R. M.; LORD, A. E.; AND HALSE, Y. H., 1988, Long-term durability and aging of geotextiles: Geotextiles and Geomembranes, Vol. 1, No. 1 2, pp MIURA, N. AND CHAI, J. C., 2000, Discharge capacity of prefabricated vertical drains confined in clay: Geosynthetics International, Vol. 7, No. 2, pp NILEX CORPORATION, 2009, Wick Drain MD-7407 Product Specifications: Electronic document (accessed October 2009), available at 20Wick%20Drain%20MD-7407.pdf PALMEIRA, E. M. AND GARDONI, M. G., 2000, The influence of partial clogging and pressure on the behavior of geotextiles in drainage systems: Geosynthetics International, Vol. 7, No. 4 6, pp ROLLIN, A. L. AND LOMBARD, G., 1988, Mechanisms affecting long-term filtration behavior of geotextiles: Geotextiles and Geomembranes, Vol. 7, No. 1 2, pp ROYSTER, D. L., 1980, Horizontal Drains and Horizontal Drilling: An Overview: National Academy of Sciences, Transportation Research Board Transportation Research Record 783, pp SANTI, P. M., 1999, Horizontal wick drains as a cost effective method to stabilize landslides: Geotechnical News, Vol. 17, No. 2, pp SANTI, P. M.; CRENSHAW, B. A.; AND ELIFRITS, C. D., 2003, Demonstration projects using wick drains to stabilize landslides: Environmental & Engineering Geoscience, Vol. 9, No. 4, pp SANTI, P. M. AND ELIFRITS, C. D., 2002, Landslide Stabilization Using Wick Drains: Unpublished Project Report prepared for the NCHRP-IDEA Project Committee, Transportation Research Board, 9 p. SANTI, P. M.; ELIFRITS, C. D.; AND LILJEGREN, J. A., 2001a, Draining in a new direction: Civil Engineering, Vol. 71, No. 6, pp. A10 A16. SANTI, P. M.; ELIFRITS, C. D.; AND LILJEGREN, J. A., 2001b, Design and installation of horizontal wick drains for landslide stabilization: Transportation Research Record 1757, Paper , Transportation Research Board, National Academy of Sciences, pp SMITH, T. W. AND STAFFORD, G. V., 1957, Horizontal drains on California highways: Journal Soil Mechanics and Foundations Division, ASCE, Vol. 110, No. 11, pp WHITE, F. M., 2008, Fluid Mechanics, 6th ed.: McGraw Hill, New York, 864 p. Environmental & Engineering Geoscience, Vol. XVII, No. 2, May 2011, pp